AbstractINTRODUCTIONStudies have suggested sleep duration may be associated with Alzheimer’s disease risk, but findings based on self-reported sleep duration are likely to be influenced by reverse causation and residual confounding bias.METHODSA polygenic score (PGS) for device-measured sleep duration was constructed using LDpred2-auto in 77,770 white British UK Biobank participants. We applied the PGS to 264,746 white British participants independent of the sample from which the PGS was developed. We assessed the association of fifths of genetically predicted sleep duration with Alzheimer’s disease risk (1,451 cases/264,746 individuals over median 12.5y of follow-up).RESULTSThe PGS explained ∼2% of variation in device-measured sleep duration. Compared to individuals in the middle fifth of PGS, those in the highest fifth (indicating ∼15 mins/day longer sleep) had a lower risk of Alzheimer’s disease (HR=0.79[95%CI,0.67-0.94]).DISCUSSIONOur results indicate that genetic predisposition to relatively long sleep duration is associated with a lower Alzheimer’s disease risk.